Purification and Identification of a Tissue-specific Repressor Involved in Serum Amyloid A1 Gene Expression*

We have previously demonstrated that the 5′-flanking regions from the rat serum amyloid A1 (SAA1) promoter are necessary and sufficient to confer specific cytokine-induced expression in cultured hepatoma cells. Deletion analysis identified a tissue-specific repressor (TSR) regulatory element, located between bp −289 and −256, that functioned as a silencer, contributing to the transcription repression on SAA1 promoter in nonliver cells. When this 34-base pair TSR-binding element was used as a probe in electrophoretic mobility shift assays, an intense DNA-protein complex was detected in nuclear extracts from HeLa and several other nonliver tissues. This TSR binding activity, however, was undetectable in extracts from liver or liver-derived cells. The distribution of TSR binding activity is therefore consistent with its regulatory role in repressing SAA1 expression in a tissue-specific manner. In this study, we purified TSR protein from HeLa nuclear extracts and showed that it has a molecular mass of approximately 50 kDa. Surprisingly, protein sequencing and antibody supershift experiments identified TSR as transcription factor AP-2. Subsequent functional analysis showed that forced expression of AP-2 in HepG2 cells could indeed inhibit conditioned medium-induced SAA1 promoter activation. Moreover, expression of a dominant-negative mutant of AP-2 in HeLa cells or mutation of the AP-2-binding site led to derepression of the SAA1 promoter, presumably by neutralizing the inhibitory effects of the endogenous wild-type AP-2. Our results therefore demonstrate a novel function for AP-2 in the transcriptional repression of SAA1 promoter. Together with its tissue distribution, AP-2 may contribute to SAA1's highly liver-specific expression pattern by restricting its expression in nonliver cells.

We have previously demonstrated that the 5-flanking regions from the rat serum amyloid A1 (SAA1) promoter are necessary and sufficient to confer specific cytokineinduced expression in cultured hepatoma cells. Deletion analysis identified a tissue-specific repressor (TSR) regulatory element, located between bp ؊289 and ؊256, that functioned as a silencer, contributing to the transcription repression on SAA1 promoter in nonliver cells. When this 34-base pair TSR-binding element was used as a probe in electrophoretic mobility shift assays, an intense DNA-protein complex was detected in nuclear extracts from HeLa and several other nonliver tissues. This TSR binding activity, however, was undetectable in extracts from liver or liver-derived cells. The distribution of TSR binding activity is therefore consistent with its regulatory role in repressing SAA1 expression in a tissue-specific manner. In this study, we purified TSR protein from HeLa nuclear extracts and showed that it has a molecular mass of approximately 50 kDa. Surprisingly, protein sequencing and antibody supershift experiments identified TSR as transcription factor AP-2. Subsequent functional analysis showed that forced expression of AP-2 in HepG2 cells could indeed inhibit conditioned medium-induced SAA1 promoter activation. Moreover, expression of a dominant-negative mutant of AP-2 in HeLa cells or mutation of the AP-2-binding site led to derepression of the SAA1 promoter, presumably by neutralizing the inhibitory effects of the endogenous wild-type AP-2. Our results therefore demonstrate a novel function for AP-2 in the transcriptional repression of SAA1 promoter. Together with its tissue distribution, AP-2 may contribute to SAA1's highly liverspecific expression pattern by restricting its expression in nonliver cells.
Multicellular organisms are composed of hundreds of distinct, highly specialized cell types, each of which requires the expression of a unique assortment of genes. Studies on the regulation of cell type-specific genes have led to the identification of both positive and negative regulatory sequences, located mostly at their 5Ј-flanking regions (1)(2)(3)(4)(5). While positive control mechanisms are essential in up-regulating the expression of eukaryotic genes, it has become increasingly clear that active repression is also critical and often contributes significantly to the proper spatial and temporal expression of tissue-specific genes. One mechanism of negative regulation involves selective transcription repression through sequence-specific DNA-binding proteins. These negative regulators could inhibit the expression of a specific gene by interfering with any of the several steps in the transcription initiation pathway (6 -8), such as by competing with transactivators for specific DNA-binding sites (9,10) or by interfering with the activity of a DNA-bound activator by quenching or masking its activation potential (11,12). In addition, negative regulators might affect the function of transcription machinery by interfering with the general transcription factors and the assembly or disassembly of the pre-initiation complexes (13,14). A number of positive-acting transcription factors that regulate multiple lineage-specific target genes have been shown to function as master regulators for cell type determination or differentiation (15)(16)(17). Likewise, similar master negative regulators that modify the expression of cell type-specific genes have been described (18 -22).
Regulation of acute-phase genes in hepatocytes by the proinflammatory cytokines interleukin-1 (IL-1), 1 IL-6, and tumor necrosis factor has been studied (23,24). One of the major acute-phase proteins is serum amyloid A (SAA) (25,26). Its serum concentration increases 1,000-fold following acute inflammation, being regulated primarily by the 200 -300-fold increase in SAA gene transcription (27). In mice, the SAA gene family consists of four genes (SAA1, SAA2, SAA3, and SAA5) and a pseudogene (28). Whereas expression of SAA1, SAA2, and SAA3 is dramatically induced following inflammation and each contributes equally to the increased SAA mRNA levels in the liver, SAA5 expression is induced to a much lower level and with different induction kinetics (29). In addition to differences in their response to inflammatory cytokines, the SAA genes also differ in their tissue distribution. While SAA3 is expressed in the liver and a few other tissues, including the macrophages and intestine, the expression of SAA1 and SAA2 is highly cell type-specific, restricting exclusively to the liver hepatocytes (27,30,31). The distinct expression patterns of SAA genes represent excellent model systems for studying not only their cytokine-induced regulation but also their liver-specific expression.
cis-Regulatory elements containing the NFB-and C/EBPbinding sites have been identified in the human SAA1 promoter and shown to be necessary for its cytokine responsiveness (32,33). Studies with the mouse SAA3 promoter demonstrated that a 350-bp promoter fragment could confer cytokine-induced response (34). Within this promoter fragment, a distal response element that contains binding sites for C/EBP and a novel transcription factor SEF has been identified and shown to possess properties of an inducible transcription enhancer (35). In the rat SAA1 promoter, a 300-bp region has been shown to be necessary and sufficient to confer both cytokine-responsive and liver-specific regulation in cultured hepatoma cells (10,36). Functional analyses have demonstrated that a 66-bp DNA fragment spanning bp Ϫ138 to Ϫ73 could confer cytokine responsiveness to a heterologous promoter (37). Within this 66-bp cytokine response unit reside binding sites for the transcription factors NFB, C/EBP, and YY-1. Sitespecific mutation studies indicated that NFB and C/EBP function cooperatively to induce SAA1 promoter activity in response to cytokine stimulation. YY-1, on the other hand, functions as a repressor in opposing NFB-mediated transcription activation, contributing to SAA1's low basal expression and the transience of its expression following inflammation (37). In addition to these factors binding to the cytokine response unit, an additional regulatory element that functions to repress SAA1 promoter activity has been identified (58). This element, however, differs from YY-1 in that it represses SAA1 promoter activity only in HeLa (nonliver) cells and has no inhibitory activity in HepG2 (liver-derived) cells (36,37). Consistent with its potential role as a tissue-specific repressor, this regulatory DNA element specifically interacts with a nuclear factor that is abundant in HeLa, placenta, and several other nonliver cell lines but is absent in the liver (58). This nuclear factor is thus termed tissue-specific repressor (TSR).
To investigate further its role in SAA1 gene transcription, we report here the purification, partial amino acid sequence, and the identity of TSR. Amino acid sequence data and antibody supershift analysis revealed that TSR is identical to transcription factor AP-2␣-1. Subsequent functional analysis demonstrated that, when expressed in HepG2 cells, AP-2 can indeed inhibit cytokine-induced SAA1 promoter activation. Consistent with its repressor function, expression of a dominant-negative mutant of AP-2 in HeLa cells resulted in derepression of SAA1 promoter. Taken together, our results suggest a novel role for AP-2 in repressing SAA1 expression in nonliver cells, which thus could contribute to its liver cell-specific expression.

EXPERIMENTAL PROCEDURES
Cell Culture and Nuclear Extract Preparation-HeLa cells were adapted to suspension growth in Spinner flasks with Joklik-modified minimum essential medium (Life Technologies, Inc.) supplemented with 5% (v/v) bovine serum (HyClone). Cells were grown to a density of 9 ϫ 10 5 cells/ml and were maintained by daily dilution with fresh culture medium to 4.5 ϫ 10 5 cells/ml. Nuclear extracts were prepared essentially as described (38). Briefly, HeLa cell pellets were washed twice with ice-cold phosphate-buffered saline and resuspended in 5 cell volumes of hypotonic buffer (10 mM HEPES, pH 7.6, 1.5 mM MgCl 2 , 10 mM KCl) containing freshly added 0.2 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM benzamidine, and 14 mM ␤-mercaptoethanol (␤-ME). After incubation on ice for 10 min, cells were lysed in a Dounce homogenizer (20 up-and-down strokes). Nuclei were then pelleted at 3400 ϫ g for 15 min, resuspended in 3.5 packed nuclear volumes of high salt buffer (20 mM HEPES, pH 7.9, 25% glycerol, 1.5 mM MgCl 2 , 0.75 M KCl, 1 mM EDTA, 1 mM benzamidine, 0.2 mM PMSF, and 14 mM ␤-ME) and incubated on ice for 15 min. After centrifugation at 10,000 ϫ g for 30 min, the crude nuclear extracts were removed and processed as described (see below).
Preparation of Magnetic DNA Affinity Beads-The double-stranded synthetic oligonucleotides 5Ј-CTTTCACTCTATACCTCAGGCAGCTA-AGGA-3Ј and 5Ј-TCCTTAGCTGCCTGAGGTATAGAGTGAAAG-3Ј, derived from the rat SAA1 promoter (bp Ϫ285 to Ϫ256) that contains the TSR-binding site, were annealed and oligomerized by ligation using T4 DNA ligase. The ligation product was inserted into the SmaI site of the pGEM vector (Promega). One clone containing eight copies of the TSRbinding site was selected and used for the preparation of DNA affinity magnetic beads. The procedure for preparing the DNA affinity beads was carried out as described (39). Briefly, a 270-bp EcoRI to BamHI fragment harboring eight copies of TSR-recognition sequence was purified from agarose gels and end-labeled at the EcoRI site using biotin-18-dATP (Life Technologies, Inc.) and Klenow fragment of DNA polymerase. Twenty mg of prewashed Dynabeads TM M-280 streptavidin (Dynal) were mixed with 50 g of biotinylated DNA fragment in TE buffer containing 1 M NaCl and mixed gently at room temperature for 30 min. After binding, the DNA-magnetic beads were washed to remove uncoupled DNA fragments and then stored at 4°C in TE buffer containing 0.1 M NaCl. We routinely obtain DNA beads that contain 2 g of DNA/mg of magnetic beads.
Purification of TSR-HeLa nuclear extracts were diluted to 30 mM NaCl with Buffer A (25 mM Tris-HCl, pH 7.3, 10% glycerol, 1 mM benzamidine, 1 mM EDTA, 14 mM ␤-ME, 0.2 mM PMSF), and applied to a 160-ml DEAE-Sephacel column at a flow rate of 3 ml/min. TSR activity was eluted from this column with Buffer A containing 0.2 M NaCl. The DEAE eluate was directly applied to a pre-equilibrated P11 phosphocellulose column (100-ml bed volume). After the column was washed with several bed volumes of Buffer A containing 0.2 M NaCl, TSR binding activity was subsequently eluted with 0.5 M NaCl in Buffer A. After its dilution to 0.2 M NaCl with Buffer A, the TSR-containing sample was applied to a 50-ml heparin-agarose column at a flow rate of 1 ml/min. The heparin-agarose column was washed extensively with Buffer A containing 0.5 M NaCl, and the bound protein was eluted with 0.8 M NaCl in Buffer A. After concentration on phosphocellulose, the TSR-containing sample was subjected to DNA affinity chromatography. The concentrated protein sample was adjusted to 0.2 M NaCl in Buffer B (20 mM Tris-HCl, pH 8.0, 10 mM EDTA, 10% glycerol, 1% Nonidet P-40, 1 mM dithiothreitol, 0.2 mM PMSF) and then mixed with poly(dI-dC) and magnetic beads coupled with multimerized TSR-binding sites. The amount of poly(dI-dC) used in these incubations depended on the amount of magnetic beads used. In general, approximately 50 g of poly(dI-dC)/mg of magnetic beads were used. The incubation mixture was placed on a roller at room temperature for 25 min before subjecting it to magnetic separation. After incubation and magnetic separation, the beads were washed twice by resuspension in Buffer B containing 0.2 M NaCl. TSR binding activity was subsequently eluted from the affinity column with Buffer B containing 0.7 M NaCl. To minimize losses of TSR due to nonspecific binding, 50 g/ml bovine serum albumin was included in the final elution buffer. The DNA affinity-purified sample was subjected to a second cycle of magnetic DNA affinity chromatography, using only 20% of the amount of magnetic beads in the first cycle.
Electrophoretic Mobility Shift Assays (EMSA)-TSR activities were monitored throughout the purification procedure by incubation of eluted protein samples with 32 P-labeled DNA fragments (2 ϫ 10 4 cpm) (35). In assays with the affinity-purified TSR preparations, 5 g of bovine serum albumin was also included in the binding reactions to minimize loss of TSR due to nonspecific binding. The gels were dried and subjected to autoradiography. Protein complex formation was quantified by PhosphorImager (Molecular Dynamics). In the oligonucleotide competition experiments, the protein samples were incubated with 32 P-labeled probe in the presence of 100-fold molar excess of wild-type or mutant competitor DNAs. The oligonucleotides used were wild-type (5Ј-AAGCTTCTCTATACCTCAGGCAGCT-3Ј) and mutant (5Ј-AAGCTTCTCTATACCTTAAGCAGCT-3Ј) TSR-binding sequences and the consensus wild-type (5Ј-AGGAATTGACCGCCCGCGGCCGT-GGTCAGAG-3Ј) and mutant (5Ј-AGGAACTGACCGACCGCTGCCGT-GGTCAGAG-3Ј) AP-2-binding sequences (40).
Rabbit polyclonal antibody against the C-terminal region (amino acids 420 -437) of human AP-2 was kindly provided by Dr. M. Tainsky (40) and was used in antibody supershift experiments and Western blot analysis. In supershift experiments, DNA affinity-purified TSR was incubated with 32 P-labeled probe in the presence of anti-AP-2 antiserum (1:1200 dilution) or preimmune serum for 30 min at 4°C. The reaction mixtures were then subjected to electrophoresis as above.
SDS-PAGE, Silver Staining, and Western Blot Analysis-Protein concentrations were determined by the Bradford protein assay (41). SDS-PAGE was performed as described by Laemmli (42), and the protein bands were visualized by silver staining (43).
For Western blot analysis, DNA affinity-purified TSR (10 ng) was run on SDS-PAGE, transferred overnight onto nitrocellulose membrane (Bio-Rad), and probed with anti-human AP-2 antibody. Horseradish peroxidase-conjugated goat anti-rabbit IgG was used as secondary antibody and was detected by chemiluminescence.
UV-induced Cross-linking of Protein-DNA Complexes-UV crosslinking was performed as described (44). Briefly, end-labeled TSR probe was incubated with affinity-purified TSR samples in EMSA conditions. The reaction mixture (60 l) was then irradiated for 10 min using a UV transilluminator (254 nm, 7000 milliwatts/cm 2 ) at a distance of 4 cm from the UV source. After electrophoresis on an 8% SDS-polyacrylamide gel, the protein-DNA complexes were identified by autoradiography. The molecular mass of the protein was estimated after subtracting the mass contributed by the DNA probe.
For transient transfection assays, HepG2 or HeLa cells were seeded in 60-mm culture dishes at 1.5-2 ϫ 10 5 cells/dish. Sixteen hr later, cells were transfected with indicated plasmid DNAs using the MBS mammalian transfection kit (Stratagene). Approximately 16 -20 h after transfection, cells were stimulated with basal medium, conditioned medium (CM), or IFN␣. Cell extracts were assayed for protein content, and the CAT activity was quantified as described (37). The averages and standard errors from three or four independent experiments were calculated relative to the activities for the control samples, to which a value of 1.0 was assigned.

RESULTS
Purification of TSR-We have previously observed a regulatory element in the rat SAA1 promoter that functioned, in transient transfection experiments, as a transcription repressor element in HeLa cells but had no inhibitory effects in hepatoma HepG2 cells (36,37,58). Furthermore, when a DNA fragment from this regulatory region was used as probe in EMSA, a strong binding activity was detected in extracts from several nonliver cells but was undetectable in liver or liverderived cells (58). Because the distribution of this nuclear factor was cell-type specific and correlated with the repressive effects of this regulatory element, this nuclear factor was termed tissue-specific repressor (TSR).
To further analyze the role of TSR on SAA1 gene regulation, we sought to purify this repressor from HeLa nuclear extracts. To facilitate its purification, we initially defined and optimized conditions that would minimize protein degradation and at the same time maintain maximum TSR binding. We examined the effects of various concentrations of EDTA, NaCl, and Nonidet P-40 on the ability of TSR to bind DNA. Our results showed that TSR binding activities were at or near optimal levels under a wide range of concentrations (2-15 mM EDTA, 50 -150 mM NaCl, and 0 -4% Nonidet P-40) (data not shown). These optimum conditions for TSR binding were then incorporated into our purification scheme to maximize the efficiency, specificity, and recovery of TSR from ion-exchange chromatography columns.
TSR was purified from HeLa nuclear extracts according to the procedure described under "Experimental Procedures." TSR activity was monitored at each stage of purification by EMSA using the TSR-binding sequence as probe. Crude HeLa nuclear extracts, partially purified fractions from DEAE-Sephacel, phosphocellulose, and heparin-agarose columns and the highly purified eluate from DNA affinity magnetic beads all formed protein-DNA complexes with identical mobilities (Fig.  1). Formation of these complexes was specifically inhibited by 100-fold molar excess of unlabeled wild-type TSR-binding site but not by mutated TSR-binding oligonucleotides. The recovery of TSR protein and its binding activities after each purification step is shown in Table I. The first two steps of purification resulted in only a modest increase in the specific activity of TSR. In comparison, the heparin-agarose column resulted in a 7-fold enrichment of TSR binding activity. This is reflected by the sharp difference in the protein banding patterns displayed by SDS-PAGE ( Fig. 2A, compare lane 5 with lanes 2-4). Although the heparin-agarose chromatography step helped to enrich TSR binding activities, it was accompanied by a large loss in total TSR activity.
The heparin-agarose eluate was diluted and then subjected to DNA affinity purification. To facilitate DNA affinity purification, we examined parameters that would help to minimize nonspecific binding of contaminating proteins while at the same time maximize TSR binding. We found that by raising the NaCl concentration to 0.2 M in the binding buffer greatly reduced the amount of contaminating proteins. In addition, inclusion of 50 g/ml bovine serum albumin in the final elution buffer greatly improved the recovery of TSR from the affinity beads.
The DNA affinity purification step removed greater than 99% of the remaining contaminating protein while recovering approximately 70% of the TSR binding activity (Table I). Thus, the DNA affinity chromatography step was clearly the most significant step in TSR purification, resulting in a 90-fold purification. The affinity-purified TSR appeared to be highly purified, as judged by silver staining on SDS-PAGE with an apparent molecular mass of 50 kDa (Fig. 2A, lane 6). When the purified TSR was subjected to a second cycle of DNA affinity chromatography, the 50-kDa polypeptide was completely undetectable in the unbound fraction and was quantitatively recovered in the high salt eluate (Fig. 2A, lane 7). Although the second cycle of the DNA affinity step removed additional contaminating proteins, it only marginally improved the purity of TSR.
To confirm that the DNA affinity-purified protein is indeed TSR and exhibits its sequence binding specificities, the purified TSR was further examined by UV cross-linking experiments. The affinity-purified TSR was incubated with 32 P-labeled TSRbinding sequence for 30 min. After incubation, the mixture was irradiated with UV light to covalently link the DNA-binding polypeptide to the DNA probe. The DNA-protein complex was then resolved by SDS-PAGE and visualized by autoradiography. As shown in Fig. 2B, exposure of purified TSR and its binding sequence to UV resulted in the formation of a strong protein-DNA complex that migrated as a single band on SDS-PAGE with an adjusted molecular mass of approximately 50 kDa. Formation of this DNA-protein adduct was highly specific to the TSR-binding sequence, since it could be inhibited by an excess of unlabeled wild-type but not by mutant TSR-binding site oligonucleotides. Furthermore, no specific bands were detected without UV irradiation (data not shown).
Identification of TSR as AP-2-To determine the identity of TSR protein, the 50-kDa protein band was subjected to amino acid sequence determination. The N-terminal 20 amino acids obtained, MLWKLTDNIKYEDCEDRHDG, were found to match exactly with the first 20 amino acids for the transcription factor AP-2␣-1 (45). In addition, TSR's molecular mass and its tissue distribution (high in HeLa cells and absent in liver cells) are also consistent with it being AP-2. To compare the sequence specificity of TSR binding with that of AP-2, oligonucleotide competition experiments were carried out using purified TSR and recombinant AP-2 (rAP-2) as protein sources, and TSR-binding site and consensus AP-2-binding site oligonucleotides as competitors. As shown in Fig. 3A, the TSR-DNA complex formed can be specifically competed by the wild-type TSRbinding sequence and by the wild-type AP-2-binding sequence, but not by their respective mutant oligonucleotides. When rAP-2 was incubated with radiolabeled TSR-binding site, a complex, indistinguishable from that of TSR-DNA complex, was also formed. Similarly, this complex could be specifically competed by wild-type TSR-and AP-2-binding site oligonucleotides and not by mutant oligonucleotides (Fig. 3A). Taken together, these results indicate that TSR and AP-2 have essentially identical sequence specificity in their DNA binding.
To further determine whether TSR and AP-2 are indeed identical and are antigenically related, specific rabbit polyclonal anti-AP-2 antibodies were used in supershift experiments with purified TSR protein. As a control, parallel reac-tions were carried out with rAP-2. As shown in Fig. 3B, both purified TSR and rAP-2 formed strong protein-DNA complexes with 32 P-labeled TSR-binding site probe. Addition of anti-AP-2 antibodies, but not preimmune serum, completely supershifted both TSR-DNA and AP-2-DNA complexes. Finally, when DNA affinity-purified TSR was subjected to Western blot analysis, a single band with a molecular mass of 50 kDa was detected with anti-AP-2 antibodies (Fig. 3C). Taken together, our results demonstrate that TSR and AP-2 have indistinguishable DNA sequence-binding specificities and share common antigenic determinants, supporting the conclusion that TSR is in fact AP-2.
Overexpression of AP-2 Inhibits Cytokine-induced Activation of Rat SAA1 Promoter in Liver Cells-We initially identified TSR based on its inhibitory activity on SAA1 promoter. Subsequent amino acid sequence data and biochemical analysis led us to the conclusion that TSR is AP-2. To determine if AP-2 could exert inhibitory effects on SAA1 promoter, we cotransfected HepG2 cells with pSAA1/CAT(Ϫ304) reporter gene along with an AP-2-expression plasmid. As shown in Fig. 4A, forced expression of AP-2 in HepG2 cells inhibited SAA1 promoter activity in a dose-dependent manner. With only 0.4 g of AP-2-expression vector, CM-induced SAA1 promoter activity was repressed by more than 70%. At higher AP-2 levels, both basal and cytokine-induced CAT activities were completely inhibited. In contrast, expression of a frameshift AP-2 mutant did not exert any inhibitory effect on SAA1 promoter activity. To demonstrate that the inhibitory effects observed with AP-2 are specific to SAA1 promoter and not due to nonspecific inhibition of general transcription, two other reporter constructs were tested for AP-2's effects on their promoter activities. The 3XAP-2RE hMt /tkCAT reporter was chosen because AP-2 has been a DNA binding activities were measured by EMSA. One unit of activity is defined as the amount of protein required to shift 10% labeled DNA in our standard assay conditions.

FIG. 2. Purification of TSR from HeLa nuclear extracts.
A, SDS-PAGE analysis. Protein samples at various stages of purification were subjected to SDS-PAGE analysis and visualized by silver staining. NE, nuclear extract; DE, DEAE-Sephacel; PC, phosphocellulose; Heparin, heparin-agarose; 1st Aff, first cycle DNA affinity; 2nd Aff, second cycle DNA affinity. B, UV-induced cross-linking of protein-DNA complexes. DNA affinity-purified TSR was incubated with 32 P-labeled probe in the absence (Ϫ) or presence of 100-fold molar excess of wild-type (WT) or mutant (mt) TSR oligonucleotide competitors. After incubation, the reaction mixtures were irradiated with UV as described in "Experimental Procedures". The DNA-protein adducts were then resolved by SDS-PAGE and visualized by autoradiography. Positions of the molecular weight markers are indicated.
shown to activate rather than repress its promoter activity (40). In agreement with earlier studies, expression of AP-2 in HepG2 cells resulted in 4-fold activation of the reporter gene activity (Fig. 4B). As an additional control, AP-2 expression vector was also cotransfected with an IFN␣-responsive 6-16/ CAT reporter in which no known AP-2-binding site has been reported. When stimulated with IFN␣, there was a 6-fold increase in CAT activity and overexpression of AP-2 had no inhibitory effect on 6-16/CAT reporter gene's responsiveness to IFN␣ (Fig. 4B). These results clearly demonstrated that AP-2's inhibitory effects on SAA1/CAT activity are quite specific and are not due to general promoter inhibition as the result of AP-2 overexpression.
Derepression of the SAA1/CAT Reporter Gene in HeLa Cells-One reason that pSAA1/CAT(Ϫ304) reporter gene cannot be induced by cytokines in HeLa cells could be due to the presence of endogenous AP-2 and not due to absence of positive regulatory signals or transcription factors. To test this hypothesis, two complementary approaches were examined. We cotransfected pSAA1/CAT(Ϫ304) reporter gene with a dominantnegative mutant of AP-2 in which a portion of the DNA-binding domain has been deleted while retaining the dimerization domain. Consequently, this dominant-negative mutant can form dimers with endogenous wild-type AP-2 but the heterodimers formed are unable to bind DNA. As shown in Fig. 5A, overexpression of the dominant-negative AP-2 mutant, pAP-2⌬166 -277, in HeLa cells resulted in a dose-dependent derepression of the SAA1 promoter. To investigate whether the lack of SAA1 promoter activities in HeLa cells is, at least in part, due to repression by the endogenous AP-2 through its binding at the promoter, we generated a mutant construct in which the AP-2 site has been mutated. The effects of such mutation on SAA1 promoter's cytokine responsiveness were analyzed. As expected, the construct with wild type AP-2-binding site was nonresponsive to CM stimulation when transfected into HeLa cells. Mutation of the AP-2-binding site, however, resulted in approximately 3-fold increase in CAT activity in response to CM stimulation (Fig. 5B). This observation is reminiscent to the results obtained with the reporter construct in which the AP-2-binding site has been deleted (36,58). Taken together, these results are consistent with the notion that endogenous AP-2 in HeLa cells contributes to the inhibition of SAA1 promoter activity. When the AP-2 activity was neutralized, either by mutation of its binding site or by overexpressing the dominant-negative mutant, resulted in derepression of the SAA1 promoter in HeLa cells. DISCUSSION Studies on transcription regulation have often focused on mechanisms of transcription activation. There are, however, an increasing number of examples where transcription repression has also been shown to play an important role in regulating the expression of many cell type-specific genes (18 -22). At least two types of repressing influences may act on genes. One type of transcription repression comes from a set of repressors that inhibit transcription in a more gene-specific manner. These repressors either down-regulate the activity of one or more positive-acting transcription factors or possess intrinsic repressing activity and could inhibit transcription initiation directly. Many positive-acting transcription factors have been shown to function as master regulators in lineage-specific target genes. For instance, the MyoD protein can activate the myogenic program by binding directly to the control regions of muscle-specific genes (17). A pituitary-specific transcription factor, Pit-1, can transactivate growth hormone, prolactin, and thyroid-stimulating hormone gene promoters (15). Transcription factors that function as master negative regulators of cell-type determination or differentiation have also been reported (18 -22). One such negative regulator is the neuronrestrictive silencer factor (NRSF) that could potentially regulate a large battery of neuron-specific genes in nonneuronal tissues (18,19). Expression of NRSF mRNA was detected in most nonneuronal tissues at several developmental stages and in undifferentiated neuronal progenitors, but absent in differentiated neurons.
Our earlier studies identified a negative regulatory element FIG. 3. DNA binding properties and antigenic determinants of purified TSR and rAP-2. A, TSR-binding fragment was 32 P-labeled and used as probe in EMSA. Purified TSR and rAP-2 were incubated with labeled probe in the absence (Ϫ) or presence of 100-fold molar excess of unlabeled oligonucleotide competitors. Binding site oligonucleotides corresponding to wild-type TSR (WT-TSR), mutated TSR (mt-TSR), wild-type AP-2 (WT-AP-2), and mutated AP-2 (mt-AP-2) were used as competitors. B, purified TSR and rAP-2 were incubated with labeled TSR-binding fragment. The TSR-DNA and AP-2-DNA complexes were incubated with preimmune or anti-AP-2 antibodies before they were run on polyacrylamide gel. C, DNA affinity-purified TSR (10 ng) was run on SDS-PAGE, transferred to membrane, and probed with anti-AP-2 antibody. Position of the specific antibody-protein complex is indicated by the solid arrow.
in the rat SAA1 promoter, which, in transient transfection assays, can specifically inhibit SAA1 promoter activity in HeLa cells while has no inhibitory effect in HepG2 cells (58). This regulatory DNA element specifically interacts with a nuclear factor termed TSR that is abundant in HeLa, placenta, and several other nonliver cell lines but is absent in the liver. In this study, we purified TSR from HeLa nuclear extracts to near homogeneity by using a combination of conventional ion-exchange columns and magnetic DNA affinity beads. The most important purification steps were the heparin-agarose and DNA affinity chromatography. Binding of TSR to the heparinagarose column was unexpectedly tight, requiring 800 mM NaCl to elute TSR. This step, however, resulted in approximately 7-fold purification and, more importantly, removed majority of the contaminating proteins (Table I). Clearly, the DNA affinity chromatograhy was by far the most efficient purification step. By enriching the TSR binding activity prior to DNA affinity chromatography, and by addition of nonspecific competitor DNA to the DNA affinity beads, we achieved about 90-fold purification in a single step.
Much to our surprise, when purified TSR protein was sequenced, the amino acid sequence obtained matched perfectly with the N-terminal 20 amino acids from the human AP-2. Subsequent biochemical analyses and comparison with the known characteristics of AP-2 further confirmed that TSR is a member of the AP-2 protein family, and more specifically, iden-tical to the AP-2␣-1 isoform. Several lines of evidence support this conclusion. (a) Protein sequencing of purified TSR showed identical amino acid sequence with that of AP-2␣-1. This 20amino acid sequence distinguishes AP-2␣-1 from other AP-2␣ isoforms, i.e., AP-2␣-2, -3, and -4, which share identical Cterminal amino acid sequence (45) and from AP-2␤ (46). (b) Anti-AP-2 antibodies could specifically recognize TSR protein both in supershift experiment and in Western blot analysis. Since the anti-AP-2 antibody used in our experiments specifically recognizes the AP-2␣ isoforms but not the alternatively spliced AP-2B isoform (47), we can thus conclude that TSR is not a member of the AP-2B proteins. (c) The molecular masses for both TSR and AP-2␣-1 proteins were estimated to be 50 kDa. (d) Like AP-2, TSR expression is abundant in HeLa cells but is totally absent in liver or liver-derived cells. (e) Oligonucleotide competition experiments showed that TSR and AP-2 have indistinguishable DNA-binding specificities.
AP-2 was first isolated from HeLa cells by DNA affinity chromatography using specific binding sites from SV40 and human metallothionein IIA promoters (48,49). It is a retinoic acid-inducible and developmentally regulated trancription factor that functions mainly as an activator in the regulation of genes involved in the morphogenesis of the peripheral nervous system, face, limbs, skin, and nepheric tissues (3,50,51). However, it has been shown recently that AP-2 can also negatively regulate the transcription of type I collagen, K3 keratin, acetylcholinesterase, prothymosin, ornithine decarboxylase,

FIG. 4. Overexpression of AP-2 in HepG2 cells inhibits CMmediated induction of SAA1 promoter.
A, HepG2 cells were cotransfected with 8 g of pSAA1/CAT(Ϫ304) DNA and either wild-type AP-2-expression vector (AP-2) or frameshift AP-2 mutant (fsAP-2). Approximately 18 h after transfection, cells were treated with (ϩ) or without (Ϫ) 50% CM. Cells were harvested 16 h later for CAT assays. -Fold induction was calculated relative to the CAT activity for pSAA1/ CAT(Ϫ304) without CM treatment, to which a value of 1.0 was assigned. B, HepG2 cells were cotransfected with 4 g of 3XAP-2RE hMt / tkCAT DNA (AP-2RE hMt ) and indicated amounts of wild-type AP-2 vector (AP-2) or frameshift mutant AP-2 vector (fsAP-2). CAT assays were performed 40 h after transfection. For the 6-16/CAT reporter gene, HepG2 cells were cotransfected with 4 g of 6-16/CAT with wild-type AP-2-expression vector. Approximately 20 h after transfection, cells were treated without (Ϫ) or with (ϩ) 300 units/ml IFN␣. Results were calculated relative to the activities for the untreated reporter gene without cotransfection, to which a value of 1.0 was assigned.

FIG. 5. Derepression of SAA1 promoter activity in HeLa cells.
A, dominant-negative AP-2 mutant. pSAA1/CAT(Ϫ304) DNA was cotransfected into HeLa cells with a dominant-negative AP-2 mutant expression vector (AP-2⌬166 -277) (1 or 2 g) or empty vector. Approximately 20 h after transfection, cells were treated with (ϩ) or without (Ϫ) CM. After 20 h of treatment, cells were harvested for CAT assays. B, AP-2-binding site mutant. Wild type SAA1 promoter construct and AP-2-binding site mutant construct (mAP-2) were transfected into HeLa cells. Transfected cells were stimulated with CM for 20 h before harvested for CAT assays. CAT activities were quantified by Phospho-rImager. Results were calculated relative to the activities for the untreated reporter gene, to which a value of 1.0 was assigned. ratina fatty-acid-binding protein, and C/EBP␣ (52)(53)(54)(55)(56)(57). Therefore, AP-2 is a multifaceted transcription factor that can function either as a transcription activator or a transcription repressor, depending on the context of its target promoter. Our finding that AP-2 is involved in the negative regulation of SAA1 gene transcription in HeLa cells adds yet another dimension to its diverse cellular function. The molecular mechanisms by which AP-2 represses SAA1 promoter are unclear. However, AP-2 has been reported to block Myc-mediated transactivation by competing with Myc/Max heterodimers for DNA binding or by interacting with Myc and thus impairs its DNA binding activity (52). More recently, it has been shown that in undifferentiated corneal epithelial cells, AP-2 can compete with Sp1 for overlapping binding sites. Since Sp1 functions as a strong activator in these cells to transactivate the K3 gene promoter, reduction of its binding to the K3 gene enhancer due to AP-2 competition results in suppression of K3 gene expression (55). In the SAA1 promoter, we show that binding of AP-2 to the promoter is required for its repression effects, since either mutation of AP-2-binding site or expression of a dominantnegative AP-2 mutant resulted in derepression of SAA1 promoter activity in HeLa cells. The possibility that AP-2 may interfere, directly or indirectly, with the activities of C/EBP or NFB to repress SAA1 gene promoter in HeLa cells cannot be ruled out and is under investigation.
Given its repressive function in restricting SAA1 gene expression in HeLa cells, it raises the question whether AP-2 might similarly repress the expression of other liver-specific genes in nonliver cells. In this regard, our preliminary data indeed suggest that several other liver gene promoters may contain functional AP-2-binding sites. 2 Thus, AP-2 may play a broad functional role in regulating the expression of other liver genes and contributing to their liver cell-specific expression pattern. Studies are in progress to further elucidate the molecular mechanisms by which AP-2 represses the expression of SAA1 and perhaps other liver genes. These studies should extend our understanding of not only the tissue-specific regulation of liver genes but also the role of transcription repression in conferring cell type-specific expression in eukaryotic cells.